Tag: MCMC

New release of Systemic 2 (2.14)

I have just released a new version of Systemic 2 (2.14).  Lots of little bug fixes, together with a few features.

My favorite is the “smooth orbit plot”, which recreates plots that will appear on an upcoming planet paper (I will put proper credit here once the paper is out!). The routine takes samples of orbital elements from the output of a Markov-Chain Monte Carlo or bootstrap run, and plots the orbits with some transparency, so that it is visually evident where orbits “crowd up”.

Smooth orbit plot
Smooth orbit plot made with Systemic 2.14

On top of the samples, I plotted the “best-fit” orbit in red (which hopefully will fall on top of the range of possible orbital elements!). This plot can also give some visual sense of multi-modal or non-symmetrical element distributions.

Download Systemic 2.14

Other changes:
– Added a feature for smoother/faster plotting in GUI
– Periodograms now print out the strongest peaks
– Periodograms are zoomable
– Fixed Cross-validation for fits with only one planet
– New menu item to add random Gaussian noise to data
– Smooth orbits plot from an error estimation object
– FIXED: Improved Linux installation instructions (credit for reporting: Thomas Kosvic)
– FIXED: bug in SWIFTRMVS on 32-bit installations (credit for reporting: Thomas Kosvic)

Systemic 2 Cookbook: Markov-Chain Monte Carlo

This post is the second in the “Systemic Cookbook” series, which will show how to use Systemic 2 for scientific purposes (especially its more advanced features not readily accessible from its user interface). More comprehensive documentation for Systemic is forthcoming.

All posts in this series will be collected in the “Systemic 2″ category.

I wrote a short draft guide to doing Markov-Chain Monte Carlo-type of fitting and error analysis with Systemic2. This guide covers the very basics of the algorithm, the input parameters, and a step-by-step guide on how to drive the algorithm from the user interface, or from the C library. Please note that more sophisticated MCMC algorithms are available as R packages on CRAN.

→ Download PDF